Gavin Brown
Scholar

Gavin Brown

Google Scholar ID: gBkMDloAAAAJ
University of Wisconsin–Madison
Data privacymachine learning
Homepage↗Google Scholar↗
Citations & Impact
All-time
Citations
434
 
H-index
7
 
i10-index
6
 
Publications
15
 
Co-authors
0
 
Contact
Emailgavin.brown@wisc.eduCVOpen ↗
Publications
15 items
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
  • - Publications:
  • - Tukey Depth Mechanisms for Practical Private Mean Estimation (with Lydia Zakynthinou)
  • - Insufficient Statistics Perturbation: Stable Estimators for Private Least Squares (with Jonathan Hayase, Samuel Hopkins, etc.)
  • - Private Gradient Descent for Linear Regression: Tighter Error Bounds and Instance-Specific Uncertainty Estimation (with Krishnamurthy Dvijotham, Georgina Evans, etc.)
  • - Metalearning with Very Few Samples Per Task (with Maryam Aliakbarpour, Konstantina Bairaktari, etc.)
  • - Fast, Sample-Efficient, Affine-Invariant Private Mean and Covariance Estimation for Subgaussian Distributions (with Samuel B. Hopkins, Adam Smith), COLT 2023 Best Student Paper Award
  • - Strong Memory Lower Bounds for Learning Natural Models (with Mark Bun, Adam Smith)
  • - Performative Prediction in a Stateful World (with Iden Kalemaj, Shlomi Hod)
  • - Covariance-Aware Private Mean Estimation Without Private Covariance Estimation (with Marco Gaboardi, Adam Smith, etc.), NeurIPS 2021 Spotlight Presentation
  • - When Is Memorization of Irrelevant Training Data Necessary for High-Accuracy Learning? (with Mark Bun, Vitaly Feldman, etc.)
  • - Awards:
  • - COLT 2023 Best Student Paper Award
  • - Boston University Department of Computer Science Research Excellence Award for 2022/23
  • - Teaching Fellow Excellence Award from BU's Computer Science Department
Research Experience
  • - Assistant Professor, Department of Computer Sciences, University of Wisconsin–Madison
  • - Postdoctoral Researcher, University of Washington
Education
  • - Ph.D., Boston University, Advisor: Adam Smith
  • - Postdoc, University of Washington, Advisor: Sewoong Oh
Background
  • Research Interests: Machine learning and data privacy. Focused on understanding when and why machine learning models memorize large amounts of training examples, and designing algorithms through the lens of differential privacy to address fundamental statistical problems.
Co-authors
0 total
Co-authors: 0 (list not available)